constructing graphql schema using reflection

I'm always shocked by the boilerplate I need to write to add simple structs to my schema.

What I did is: I started to write some helper methods to analyze my structs via reflection and generate the gql types from my go structs.

Looks like this:
```
type Model struct {
Id int64 `gorm:"type:bigserial;primary_key" json:"id"`
CreatedAt time.Time `sql:"DEFAULT:NOW()" json:"createdAt"`
UpdatedAt time.Time `json:"updatedAt"`
}

type IntegrationHttp struct {
Model
App *model.AppInstance `gorm:"not null;foreignkey:id;association_foreignkey:AppId" json:"app"`
AppId int64 `gorm:"not null" json:"appId"`

Method string `json:"method"`
Url string `json:"url"`
Skipped string `json:"-"`
}

func TestNewTypeFromStruct(t *testing.T) {
conf := schema.NewTypeFromStruct(IntegrationHttp{})

assert.Equal(t, "IntegrationHttp", conf.Name())

confFields := conf.Fields()
assert.Equal(t, 7, len(confFields))
}
```

Yet it takes the fields names from the `json` tag, could be configurable as well.
I also added a custom tag `gql` where you can override the type of a field with e.g. `gql:"type:String"`.

I can Imagine a bunch of more features. Since the code runs only once at startup time, I do not see any performance issues but reducing a lot of boilerplate code.

Now my questions are:
* **Is this kind of struct generation interesting for this library?**
* **Is it interesting in general (I could start another repository)?**

Or do you see any general issues with that and I should keep it private? Else I would love to share the implementation and would be happy to further improve it.